RFR: 8365911: AArch64: Fix encoding error in sve_cpy for negative floats

erifan duke at openjdk.org
Wed Aug 27 01:40:18 UTC 2025


The sve_cpy instruction is not correctly implemented for negative floating-point values. The issues include:

1. When a negative floating-point number (e.g. `-1.0`) is passed, the `checked_cast<int8_t>(pack(d))` check fails. For example, assume `d = -1.0`:
- `pack(-1.0)` returns an unsigned int with the 7th bit set, i.e., `0xf0`.
- `checked_cast<int8_t>(0xf0)` casts `0xf0` to an int8_t value, which is `-16`.
- Casting this int8_t `-16` back to unsigned int results in `0xfffffff0`.
- The check compares `0xf0` to `0xfffffff0`, which obviously fails.

2. Additionally, the encoding of the negative floating-point number is incorrect:
- The imm8 field can fall outside the valid range of **[-128, 127]**.
- Bit **13** should be encoded as **0** for floating-point numbers.

This PR fixes these issues and renames floating-point `sve_cpy` as `sve_fcpy`.

Some test cases are added to aarch64-asmtest.py, and all tests passed.

-------------

Commit messages:
 - 8365911: AArch64: Fix encoding error in sve_cpy for negative floats

Changes: https://git.openjdk.org/jdk/pull/26951/files
  Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=26951&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8365911
  Stats: 146 lines in 4 files changed: 10 ins; 6 del; 130 mod
  Patch: https://git.openjdk.org/jdk/pull/26951.diff
  Fetch: git fetch https://git.openjdk.org/jdk.git pull/26951/head:pull/26951

PR: https://git.openjdk.org/jdk/pull/26951


More information about the hotspot-compiler-dev mailing list